Spectral-Differential Feature Matching and Clustering for Multi-body Motion Estimation

System for estimating the motion of independently moving objects observed by a moving camera is presented. It consists of feature matching and multi-body motion estimating modules. Novel set of invariant features is proposed on the base of phase spectrum differentiation without information loss. Clustering the feature points and estimating the transformation model for each cluster are guided by criterion derived from the minimum description length principle that results in correct selection of number of clusters and family of transformations, as well as rejection of outliers.

[1]  Bernhard P. Wrobel,et al.  Multiple View Geometry in Computer Vision , 2001 .

[2]  P. Torr.  Geometric motion segmentation and model selection , 1998, Philosophical Transactions of the Royal Society of London. Series A: Mathematical, Physical and Engineering Sciences.

[3]  J. Rissanen,et al.  Modeling By Shortest Data Description* , 1978, Autom..

[4]  Konrad Schindler,et al.  Perspective n-View Multibody Structure-and-Motion Through Model Selection , 2006, ECCV.

[5]  S. Shankar Sastry,et al.  Two-View Multibody Structure from Motion , 2005, International Journal of Computer Vision.

[6]  A. S. Potapov,et al.  Construction of systems of local invariant image indicators based on the Fourier-Mellin transform , 2010 .

[7]  Christopher Hunt,et al.  Notes on the OpenSURF Library , 2009 .